DOSSIER PER ENGAGEMENT
Education conformity file

EDUCATION AI

Procurement-grade data and buildfor every education-AI team

One partner for training data, evaluation, and managed builds across the full range of education-AI builders. The Article 10 evidence dossier ships with every engagement.

EU AI Act Art. 10 GDPR Art. 6 / 9 COPPA FERPA WCAG 2.2 AA
Ed-tech with an ML team Read the data brief Ed-tech without an ML team Read the build brief

150+ LANGUAGES. 40,000+ CONTRIBUTORS.

SCOPE 11 SEGMENTS
Who builds education AI

WHO BUILDS EDUCATION AI

Eleven builder segments, one data and build layer

Education AI is not one buyer. These are the teams that source training data, evaluation, and managed builds from YPAI, grouped by how they buy.

Speed and volume

  • Edu-AI startups Founder / CTO One to four week cycles; bespoke collection and RLHF as they scale.
  • Language-learning AI Product lead Multilingual speech and pronunciation data, preference data across languages.
  • Tutoring AI Product lead Annotation and pedagogical evaluation for Socratic-style tutoring.

Publishers and platforms

  • Curriculum and content AI vendors Head of content AI Subject-matter annotation and content validation against standards.
  • Adaptive-learning AI ML engineering lead Interaction-trace and knowledge-component annotation.
  • Assessment AI Assessment AI lead Item evaluation, bias checks, and Annex III conformity evidence.

Institutional and enterprise

  • Higher-ed AI AI program lead Evaluation and compliance evidence for high-stakes use.
  • Corporate learning AI Head of learning Domain preference data and content evaluation by instructional experts.
  • Administrative and operations AI vendors IT or product lead Provenance and deletion controls for procurement review.

Compliance-critical

  • Accessibility and special-education AI Accessibility AI lead WCAG 2.2 AA conformance and assistive-tech evaluation.
  • Proctoring and exam-integrity AI Integrity AI lead Annex III high-risk evidence for exam monitoring.
PROCURE 4 CATEGORIES
What builders procure

WHAT BUILDERS PROCURE

Four capabilities every education-AI team buys

The categories education-AI builders procure most often, each tied to a YPAI capability and the evidence that ships with it.

  1. Annotation and RLHF

    What builders procure Text and multimodal annotation, preference data for alignment
    YPAI capability Self-hosted annotation on CVAT and Label Studio, credentialed reviewers
    Proof anchor Per-task QA, inter-rater agreement

  2. Evaluation

    What builders procure Model and content evaluation, red-team safety review
    YPAI capability Subject-matter reviewer network across domains
    Proof anchor Documented rubrics, reviewer credentials

  3. Multilingual speech collection

    What builders procure Consented audio across many languages and accents
    YPAI capability 150+ languages, consent recorded per contributor
    Proof anchor Consent ledger, GDPR Article 7

  4. EU AI Act Annex III conformity

    What builders procure Trustworthy AI evidence for high-risk education AI
    YPAI capability Article 10 evidence pack with full provenance
    Proof anchor Article 10 pack, DPIA template
WORKFORCE IDENTITY-VERIFIED
Credentialed reviewer network

THE DIFFERENTIATOR

Credentialed reviewers, not a general-purpose crowd

Education-AI builders need reviewers with real credentials for preference data and content evaluation. General-purpose annotation marketplaces do not source this workforce at education-domain granularity. The identity-verified contributor network does.

  • Licensed educators and curriculum specialists

    Standards-aligned content review

  • Subject-matter PhDs across STEM and the humanities

    Domain-accurate evaluation and preference data

  • Assessment and accessibility specialists

    Item-bias review, WCAG 2.2 AA

  • Native speakers across 150+ languages

    Language and dialect coverage

ART. 10 PRE-CONTRACT
Five regulatory controls

Five controls every education-AI deployment clears

  1. Art. 10(2)(a-g) EU / 27 MEMBER STATES

    EU AI Act

    Training, validation, and test datasets carry traceable documentation per subclause: relevant design choices (a), data origin and provenance (b), data labelling procedures (c), data preparation (d), assumptions about what the data should measure (e), bias examination in view of health and safety (f), and gaps that prevent compliance (g).

    Article 10 evidence pack PDF + JSON-LD provenance
  2. Art. 6(1)(a) / Art. 9 / Rec. 38 EU / EEA

    GDPR

    Lawful basis recorded per contributor and held in a per-contributor consent ledger. DPIA artifact filed for special-category processing where the dataset requires it. Erasure honoured within the documented SLA.

    GDPR DPIA template YAML + PDF, pre-filled
  3. 16 CFR Part 312 US (FEDERAL)

    COPPA

    Knowledge-of-age handling documented per data source, with consent collection scoped to the lawful basis. Data provenance recorded so downstream training use stays inside the consented scope.

    COPPA handling record PDF + state matrix
  4. 20 USC 1232g / 34 CFR Part 99 US (FEDERAL + 50 STATES)

    FERPA

    Educational-record disclosure controls per institution. Directory-information opt-out honoured at ingest. Disclosure-audit trail with date, recipient, and purpose retained for the legal minimum across the cohort lifecycle.

    FERPA controls map PDF, 9 pages
  5. AA conformance + EN 301 549 GLOBAL / W3C

    WCAG 2.2 AA

    Education-AI surfaces meet AA across screen-reader compatibility, keyboard-only navigation, focus visibility, target sizes, and assistive-tech pathways. Conformance statement on file, refreshed at every release.

    WCAG conformance map PDF + WAI-ARIA notes
ROUTING 2 ROUTES
Data route or build route

For ed-tech with an ML team

Audit-defensible training data for your education models

Per-contributor consent, Article 10 evidence, and jurisdiction maps shipped with the dataset. Your ML team integrates; YPAI carries the data dossier.

Read the data brief

For ed-tech without an ML team

Ship a regulated education-AI feature with a managed build engagement

Managed engagement from scope to handoff. Audit artifacts shipped at delivery. Your product team owns the surface; YPAI owns the model and the dossier.

Read the build brief
COVERAGE 150+ / 50+
Languages and contributors

EDUCATION PROOF / 2025

Where the evidence comes from

150+

LANGUAGES SUPPORTED

40,000+

CONTRIBUTORS

5

FRAMEWORKS ANCHORED

50+

COUNTRIES

WITNESS ROLE-ONLY
Audit-defensibility, attested

The Article 10 mapping was the difference. Two other vendors handed us a dataset. YPAI handed us a dataset we could defend in an audit, clause by clause, contributor by contributor, jurisdiction by jurisdiction.

Data Protection Officer EU regional university, education-AI procurement (role-only per customer request)
LEDGER PRE-CONTRACT
Evidence pack manifest

EVIDENCE / LEDGER

What ships with every education engagement

ypai-education-engagement/

  1. article-10-evidence.pdf EU AI Act Article 10 evidence pack EU AI ACT PRE-CONTRACT
  2. consent-ledger.pdf Per-contributor consent ledger GDPR ART. 7 PRE-CONTRACT
  3. jurisdiction-map.pdf Jurisdictional lawful-basis map 50+ JURIS. PRE-CONTRACT
  4. dpia-template.pdf Data Protection Impact Assessment GDPR PRE-CONTRACT
  5. wcag-conformance.pdf WCAG 2.2 AA conformance statement BUILD ONLY ON-REQUEST
See the full evidence ledger
NEXT ON REQUEST
Scope the engagement

Scope a governed AI data project.

Both routes start with one thirty-minute scoping call: your data, your jurisdiction, the dossier you will need to defend the deployment.

Pick a route